Search This Blog


Polling puts LA governor job on knife's edge

The 2019 Louisiana gubernatorial runoff race is closer than polling even thinks.

Since the general election put incumbent Democrat Gov. John Bel Edwards and Republican challenger Eddie Rispone through to a runoff, two polls independent polls have come out taking the temperature of the race. We Ask America had the race tied at 47-all percent, while JMC Analytics gave Edwards a 48-46 lead.

That’s close, but when breaking down each poll’s assumptions, things actually tighten more. As a reference point, keep in mind that on Oct. 12 the electorate was 69.1 percent white, 27.6 percent black, and 3.3 other race; and 44.9 percent Democrat, 37.1 percent Republican, and 18 percent other/no party. Further, females comprised 55 percent of the registered electorate eligible to vote (the state doesn’t keep statistics on proportion of the electorate voting by sex).

In comparison to 2015 turnout rates, white Democrats voted at about the same just under a 14 percentage point higher rate than blacks, but Republicans increased their turnout rate over blacks by 4 points and erased about a 3.5 point gap with white Democrats to overtake them by about a point. Back then, the 2015 runoff saw higher black turnout half offset by lower white Democrat turnout and most of the rest of that gap compensated for by slightly increased GOP turnout. This produced a small boost overall for Edwards in the runoff compared to general election.

Unfortunately for him, 2019 dynamics are somewhat different. As the 2019 general election turnout demonstrated, Republicans seem more motivated that after a dispirited intra-party battle in 2015 left them with a surviving candidate a critical mass didn’t like. Thus, at best expect no bonus for Edwards, and perhaps even a slight advantage to Rispone in terms of runoff turnout.

Now reviewing each poll’s sample:

WEA – 67 percent white, 28 percent black, 5 percent other; 41 percent registered Democrats, 37 percent registered Republicans, 19 percent other/no party; 53 percent female, 47 percent male; 70 percent contacted by landline, 30 percent by mobile.

JMC – 69 percent white, 29 percent black, 3 percent other; 45 percent registered Democrats, 35 percent registered Republicans, 20 percent other/no party; 55 percent female, 47 percent male; 65 percent by landline, 35 percent by mobile.

Given the relative proportions of the electorate in just over three weeks likely will look pretty close to what they were earlier this month, WEA undershoots on whites and JMC overshoots on blacks; and WEA undershoots on Democrats (by implication whites) and JMC undershoots on Republicans. Each shortchanges Rispone a bit (sifting through the JMC numbers implies that Rispone does a little better than Edwards with white Democrats).

Sex and phone contact source, when adjusted, also help Rispone marginally. WEA does a better job with this demographic, for in 2016 elections nationally the gap between the sexes favored by only 5.6 points. Adjusting for the differing racial proportions (only 3 points among whites but close to 10 for blacks) in a Louisiana electorate with a black proportion 2.5 times the national number, WEA goes just a bit under while JMC’s is still much higher. Consider as well anecdotally Louisiana’s political culture probably produces a smaller female/male gap than does the national ratio, which (according to JMC numbers) favors Rispone.

Phone contact source is somewhat of a shot in the dark, with the knowledge that cell phone users disproportionately are more likely to support Democrats (JMC data also implies this) and less likely to vote. Interestingly, JMC increased its proportion from its surveys prior to the general election. Finally, each does seem to weigh most heavily towards frequent voters (although neither released adequate information to define specifically how that is determined).

Sum it all up, and Rispone probably under-polls a fraction. That means, given these results, that the race teeters on a knife’s edge and could go either way, validating what one of the pollsters concluded (trite, but true) in that it all will come down to turnout.

No comments: